Kodiak AI, Inc. Q3 FY2025 Earnings Call

Thank you for joining us for Kodiak's Third Quarter 2025 Earnings Conference Call. I will now turn the call over to Lauren Harper, Kodiak's Chief of Staff. Please proceed.

Thank you, and welcome, everyone, to Kodiak's Third Quarter 2025 Earnings Call. On the call today are Don Burnette, Founder and Chief Executive Officer of Kodiak; and Surajit Datta, Chief Financial Officer of Kodiak. Our press release and an earnings presentation were issued earlier today and are posted on the Investor Relations section of our website. This call is being broadcast live via a webcast, and a replay will be available on our website after the call. Before we begin, I would like to remind you that during today's call, Kodiak will be making forward-looking statements within the meaning of the federal securities laws about financial performance and future events, including our guidance for fiscal fourth quarter and full fiscal year 2025, as well as our long-term goals. Such forward-looking statements are made pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Any statement made during this call that is not a statement of historical facts should be deemed to be a forward-looking statement. All forward-looking statements, including statements regarding our guidance for fiscal fourth quarter and full fiscal year 2025, our estimated total addressable market, our operational and product roadmap, our relationships with partners and suppliers, our ability to produce and deploy the Kodiak driver at scale, including the timing of launching driverless trucks for long-haul highway operations, our expansion plans and opportunities, and our expectations regarding future business and financial performance, including future cash flows and our path to profitability, are based upon management's current estimates and various assumptions. These statements involve material risks and uncertainties that could cause actual results or events to materially differ from those anticipated or implied by these forward-looking statements. Accordingly, you should not place undue reliance on these statements. For a list and description of the risks and uncertainties associated with our business, please see our filings with the Securities and Exchange Commission. We disclaim any obligation, except as required by law, to update or revise any financial or operational projections or other forward-looking statements, whether because of new information, future events or otherwise. Any forward-looking statements made on this call speak only as of the date of this call. Further, in addition to discussing results that are calculated in accordance with generally accepted accounting principles, we may also refer to certain non-GAAP financial measures. For more detailed information on our non-GAAP financial disclosures, including reconciliations to most comparable GAAP measures, please refer to our earnings release, which can be found on our Investor Relations website. Our discussion today also includes references to forward-looking free cash flow. Such forward-looking financial measure is provided on a non-GAAP basis without a reconciliation to the most directly comparable GAAP measure due to the inherent difficulty in forecasting and quantifying certain amounts that are necessary for such reconciliation. I will now turn the call over to Don. Please go ahead.

Thanks, Lauren, and welcome to our first earnings call as a public company. Today, we're excited to share our third quarter results. But first, I'd like to introduce you to Kodiak and talk about our industry-leading technology, our strategy, and why we believe we are positioned to capture the tremendous opportunity in the global trucking market. Kodiak delivers AI-powered driving technology that tackles some of the world's toughest driving jobs across vehicles and environments. We are a leading provider of autonomous trucking technology with 10 driverless trucks in customer operation with no human in the cab. These 10 trucks are the first to be delivered as we fulfill the world's largest known driverless trucking contract, a binding order to deploy our technology in 100 customer-owned trucks. Our differentiated technology allows us to seamlessly operate across a variety of environments, giving us the flexibility we need to focus on three large verticals: long-haul trucking, industrial trucking, and defense. Across these verticals, we work with industry-leading customers such as J.B. Hunt, Werner, C.R. England, Martin Brower, Atlas Energy Solutions, and the U.S. Army. Kodiak's core business is based on the Driver-as-a-Service, or DaaS, model, which is designed to replicate how customers pay their drivers today, either by the mile or by the vehicle. Our customers own and operate their own driverless trucks and pay us a recurring fee to utilize the Kodiak driver in their fleets. This DaaS model allows us to generate recurring revenue while keeping our balance sheet relatively asset-light. We initially launched this model with Atlas Energy Solutions, a leading logistics provider in the Permian Basin that is actively working to automate its supply chain. We intend to grow DaaS revenue, both as Atlas grows its fleet of Kodiak-powered trucks and as we expand driverless operations with new customers. I believe it is paramount to execute across three strategic pillars to launch a driverless business: technology, safety case, and product. While technology is the most visible, each of these pillars is critical to successfully launching and scaling autonomous trucks. Let me address each of these, starting with technology. The Kodiak Driver, our autonomous system combines advanced AI-powered software with modular vehicle-agnostic hardware. We designed The Kodiak Driver to operate in challenging driving environments while integrating seamlessly into our customers' fleets. This single integrated software platform is designed for deployment across three verticals: long-haul trucking, industrial trucking, and defense. It operates day and night in a wide range of weather conditions, including rain and severe dust storms and in some of the most complex scenarios in trucking. Our core technology allows the Kodiak driver to operate without high-definition maps commonly used in the AV industry. We believe this approach makes the Kodiak driver more adaptable to real-time changes on the road than traditional AVs that heavily rely on pre-maps. It also makes the Kodiak driver better able to navigate unstructured off-road environments, uniquely enabling us to pursue opportunities in the industrial and defense verticals that would otherwise be challenging if we required high-definition maps. We integrate our software into our modular autonomy hardware kit that is adaptable to a wide range of vehicle types. We've already integrated the Kodiak hardware kit into Class 8 trucks, Ford F-150s, and even treaded military vehicles. Our hardware kit includes Kodiak's proprietary sensor pods, which are self-contained modules that contain third-party cameras, radars, and LiDARs, designed to be quick and easy to swap in the field with minimal training. Our modular approach allows us to quickly install The Kodiak Driver's hardware kit into our customers' trucks, reducing the cost of deployment. Years ago, we decided that upfitting was the right go-to-market strategy to address a significant industry-wide challenge that today, none of the truck OEMs offer a driverless-ready platform. To implement this strategy, we partnered with Michigan-based Roush Industries, a leading automotive product development supplier. Roush is an industry leader in small to medium volume automotive manufacturing. Their deep experience building high-quality vehicles at scale brings efficiency and quality to our manufacturing that would be very difficult to match with in-house production. Now to our second pillar, safety. Safety is our core value at Kodiak. It is critical to building trust with customers, regulators, and the general public. The most important part of the process is building a safety case, which is a comprehensive evidence-backed argument for why The Kodiak Driver is safe to deploy. Kodiak is one of only a handful of AV companies known to have completed a driverless safety case. Developing a safety case is a massive cross-functional effort, spanning software, systems engineering, hardware, functional safety, manufacturing, operations, and even legal and policy. To help track our progress on launching The Kodiak Driver on long-haul routes, today, we are unveiling our Autonomous Readiness Measure, or ARM, which measures the percentage of claims and evidence in the Kodiak safety case for driverless operations that are materially complete. Having completed our industrial safety case, our industrial arm is necessarily at 100%. Today, I am incredibly proud to report that our long-haul arm currently stands at 78%, and we expect to make steady progress in the quarters ahead as we prepare for long-haul driverless operations in the second half of 2026. Lastly, I'd like to turn to our third pillar, the product. We take a unique customer-centric approach that has allowed us to implement our technology and build a valuable product our customers can utilize efficiently. The operational experience we have gained over seven years and 10,000 loads has enabled us to understand our customers' needs and build our products accordingly. For example, we developed our quick swap sensor pods after hearing from our customers about the importance of our hardware fitting into existing maintenance processes. Additionally, our driverless deployment with Atlas in the Permian has helped us pressure test core product features, including launching, driving, landing, and maintaining driverless vehicles. We see this generating a flywheel effect, allowing us to leverage these features and our learnings across long-haul and defense deployments. As we have already discussed, the Kodiak driver operates on highways, local streets, dirt roads, and off-road environments. This generalization enables seamless operations and shared learnings across industries. I'd like to address our unique experience and opportunity in each of these verticals. Kodiak started its journey with long-haul trucking, and this continues to be our primary focus. Given the trucking industry's $4 trillion global TAM, we see tremendous opportunity in this vertical. Today, we get paid to deliver freight from our Dallas operations hub to Houston, Oklahoma City, and Atlanta using our own trucking fleet with a safety driver. Our highly respected customers include J.B. Hunt, Werner, C.R. England, and Martin Brower. We also see a robust pipeline of customers interested in piloting and eventually deploying The Kodiak Driver in their fleets. We continue to make meaningful progress towards our goal of launching driverless operations in the long-haul vertical in the second half of next year and expect to subsequently transition our long-haul customers to our DaaS business model. The industrial trucking vertical, where we deployed our first driverless trucks includes oil and gas, mineral transportation, and logging transportation. The $68 billion global industrial market is an ideal use case for our technology. Operators in remote industrial locations face even greater difficulties recruiting and retaining drivers than long-haul carriers, meaning that autonomy can offer significant cost savings and operational efficiencies. As I mentioned previously, The Kodiak Driver's technology is well designed to work in unstructured industrial environments where we see many challenging scenarios that are less common in long-haul trucking. These include oncoming traffic, narrow uneven roads, frequent pedestrians, and occasional count. We believe that this versatility sets The Kodiak Driver's AV capabilities ahead of the competition. In addition to being a significant revenue opportunity, our industrial trucking deployment in the Permian helps us mature The Kodiak Driver across all three of our strategic pillars. The Permian is a literal and figurative sandbox for honing our technology, our safety case, and product, enabling a flywheel that is already accelerating our ability to implement our technology across verticals. The unstructured driving environment of the Permian produces more unusual edge cases compared to structured highway driving, progressing the technology through rapid learning and challenging the system. We've also gained significant product experience integrating into our customers' transportation management systems and developed processes for managing vehicle maintenance with operations personnel on the ground. We've been delivering loads with no human in the vehicle in some of the toughest weather conditions in trucking, including dust storms. This experience will enable us to provide greater value to our long-haul customers. Lastly, we're focused on adapting our virtual driver for defense purposes. The current administration has adopted a supportive posture for autonomous technology and defense use cases mirror long-haul and industrial trucking with both on-road and off-road operations. We have proven experience in defense exemplified by our $30 million contract with the U.S. Army, which we completed work on earlier in 2025. The need for autonomy in defense applications is obvious. And ultimately, we believe autonomy will be integrated into the entire Pentagon vehicle fleet. We believe Congress and the military support investments in autonomy, though the government shutdown has caused some short-term uncertainty around contracting timing. We will continue to pursue opportunities in this space going into 2026. Now turning to our Q3 results. We are pleased with our strong performance in Q3. During the third quarter, we deployed the Kodiak driver in an additional five trucks owned by our customer, Atlas Energy Solutions. We are now generating recurring revenue on a total of 10 driverless trucks equipped with the Kodiak driver, a 100% increase over Q2. As we move into 2026, we expect to accelerate our deployment as we seek to fulfill Atlas's initial order of 100 Kodiak-equipped trucks. We have also completed over 5,200 cumulative hours of paid driverless operations, which represents a 166% increase from the end of Q2. We view this as a key metric to measure the progress of an autonomous company regardless of the specific use case. Additionally, we have driven over 3 million autonomous miles and delivered over 10,000 loads for our customers, a 21% increase in cumulative deliveries over Q2. I'd like to address our recent progress by pillar. First, the technology pillar. At the beginning of Q4, we issued a new software release that, among other improvements, includes new features that allow us to reduce the need for remote assistance by 53%. This reduction, in turn, improves our ability to scale our solution. Another feature included in this new software release is a new component of our perception system that leverages generative AI-based vision language models to identify and address novel complex edge case scenarios that are rare in the real world. We believe combining our proprietary multimodal AI perception model, Giga FusionNet, with these new high-level reasoning capabilities sets a new standard for physical AI. Using this new technology, The Kodiak Driver's AI can now identify scenarios like flooded roads and car fires, rare scenarios that can be a challenge for more traditional perception techniques. This new feature allows us to better handle the long tail of complex edge cases and gives us further confidence as we move down the path towards our long-haul driverless deployment. In Q3, we announced that we have integrated NXP's ISO 26262-compliant processors and interfaces into the architecture that powers The Kodiak Driver. The addition of NXP's automotive solutions has improved The Kodiak Driver's reliability and robustness while helping us to improve vehicle uptime. Last week, we announced an expanded partnership with global Tier 1 automotive supplier, ZF. Through this expanded partnership, we purchased steering systems with redundant components for 100 Kodiak-equipped trucks. These redundant steering components are critical to ensuring our ability to safely scale The Kodiak Driver. On the safety side, we were proud to announce the results of an evaluation by Nauto, a leader in AI-powered fleet safety technology. Kodiak received the highest VERA safety score among over 1,000 fleets in Nauto's network. In fact, The Kodiak Driver received a perfect score of 100 in three out of four Nauto VERA score categories and 95 in the fourth. We believe independent safety evaluations like Nauto's help to validate what we already know: The Kodiak Driver is already among the safest drivers on American highways. Lastly, I'd like to address the product pillar. As we've discussed earlier, we continue to execute on the upfitting strategy we first articulated years ago. In Q3, Roush launched a dedicated manufacturing line to upfit trucks with Kodiak's hardware kit. With the Roush-built trucks already rolling off the line, we believe we have further solidified a leading position in building driverless trucks at scale. We also continue to strengthen our OEM relationships over the course of the quarter, and we'll continue to prioritize building those relationships. Additionally, we added the ability to haul two trailers at the same time. Pulling doubles, as it's called in trucking, is difficult for even highly skilled human drivers. In addition to extra length, the second trailer complicates every turn, making maneuvering extremely difficult. Pulling doubles is even more difficult in the Permian, which features rugged landscapes with narrow windy roads, and we achieved a significant technical milestone ahead of schedule. We also saw significant regulatory progress in Q3 and the early days of Q4. I'd like to highlight one item in particular. In early October, the U.S. Department of Transportation issued a waiver that allows AV truck operators to use flashing warning beacons as a replacement for reflective warning triangles. We integrated these warning beacons into our very first driverless-ready truck back in 2024. We are thankful for the administration's support, and we want to thank them for their leadership on this issue. Finally, over the past few months, we made tremendous progress as we prepared Kodiak to operate as a public company. We are incredibly excited for this next chapter and look forward to sharing more updates on our next earnings call. I am proud of what we have accomplished thus far and believe we are well positioned for growth. In summary, Q3 showed Kodiak executing across our three pillars: technology, safety, and product, positioning us to deliver meaningful growth and durable value for customers, partners, and shareholders. We are converting our progress into real commercial traction, including delivering additional Kodiak driver-powered trucks under the world's largest known driverless truck deployment. We head into 2026 with strong momentum across the long-haul, industrial, and defense verticals and are continuing to move toward the launch of driverless long-haul operations while scaling our industrial business. We believe these wins will allow us to grow as we remain capital efficient, providing a path to profitability and positive free cash flow in the future. I'll conclude my remarks by extending a huge thanks to all of our Kodiakers. The progress we have made is a reflection of your hard work, and I look forward to accomplishing even more with you in the future. Surajit, over to you.

Thank you, Don, and good afternoon, everyone. I am pleased to share Kodiak's financial results for the third quarter of 2025, our first as a public company. This quarter marks an important milestone for Kodiak as we continue to successfully execute our mission to accelerate adoption and commercialization of autonomous technology in a safe and reliable way. From a financial perspective, we are committed to delivering consistent value and building on the strong foundation already in place. We see potential for significant long-term growth, scale, and operating leverage. We delivered strong performance, demonstrating our ability to scale and grow the business. We also continue to focus on deploying capital efficiently, most notably by partnering with leaders in the AV and trucking ecosystems. This allows us to focus on our core competency of developing advanced AI-powered autonomy software. Revenue for the third quarter was $0.8 million. This represents a 53% growth over the prior quarter, primarily driven by an increase in Driver-as-a-Service revenue generated by our 100% quarter-over-quarter growth in customer-owned and operated driverless trucks. Let me take a moment to further explain our DaaS revenue model. Under this model, we charge our customers a single composite license fee to use The Kodiak Driver AV hardware and AI-powered autonomy software. We charge this fee on either a per vehicle or a per mile basis. This flexible pricing model is designed to align with our customers' diverse operational models. The DaaS model allows us to build predictable recurring revenue under multi-year contracts. We have already implemented the DaaS model with Atlas. In our long-haul operations, our customers currently pay us to deliver freight on Kodiak-owned autonomous trucks. We plan to transition our long-haul customers to the DaaS model once we commence our long-haul driverless operations. By integrating the Kodiak driver into customer-owned fleets, we expect to continue to build an asset-light business that scales with our customers' growth while limiting our CapEx outlay. Now turning back to the financials. GAAP operating loss for the third quarter was $30 million. Non-GAAP operating loss for the quarter, which excludes stock-based compensation, totaled $24.7 million, primarily due to continued investment in R&D and operational support for our industrial deployment. For a reconciliation of non-GAAP metrics to GAAP, please see our earnings release, which we filed prior to the call. We also incurred capital expenditures of $6.6 million, primarily to purchase AV components that we deploy on our customers' trucks. Free cash flow for the quarter was negative $40 million. This included high single-digit millions of one-time payments and public company-related costs. We ended the quarter with $146.2 million in cash and cash equivalents, including the proceeds raised as part of the de-SPAC transaction, net of fees and expenses. Turning to guidance of Q4 of fiscal year 2025. We expect to end 2025 with customer-owned and operated driverless trucks in the mid- to high teens. Q4 FY '25 free cash flow is expected to be in the range of negative $36 million to negative $38 million as we continue to invest in R&D and incur capital expenditures to purchase and deploy AV hardware on customer-owned trucks. Over the next few quarters, we expect to continue to deploy The Kodiak Driver to meet our initial contractual commitment of 100 customer-owned driverless trucks with Atlas. This is expected to drive meaningful increases in quarter-over-quarter revenue. As we look ahead, we expect our capital needs to be primarily driven by four factors: R&D investments, including safety validation; costs associated with scaling industrial commercialization; strategic initiatives to lower AV unit hardware expenditures; and public company costs. We expect that these capital needs will be partially offset by increases in DaaS revenue as well as improvements in operating leverage from scale and efficiencies. We plan to provide more detailed guidance for fiscal year 2026 as a part of our Q4 FY 2025 earnings call. We'll opportunistically seek additional financing options to strengthen our liquidity and support the next phase of growth, particularly as we build out our customer pipeline and launch driverless commercial long-haul operations in the second half of fiscal 2026. Our financial priorities remain consistent with our strategic goals that Don had laid out earlier. We want to, first, grow Driver-as-a-Service revenue with existing and new industrial and long-haul customers to build a durable recurring high-margin business model over time. Second, invest prudently in technology, safety cases, and commercial readiness to launch long-haul driverless operations in the second half of fiscal 2026. Third, implement scale and cost efficiencies into our capital-light model to achieve profitability and positive free cash flow over time. Lastly, maintain a strong balance sheet and enhance liquidity through disciplined capital planning and opportunistic financing. In summary, Kodiak is entering its next chapter with a strong foundation. Our momentum, technology leadership, and competitive position remains strong, and we are delivering high growth with focus on improving operating leverage. We are executing with fiscal discipline and transparency as we build long-term value for our customers, partners, employees, and shareholders. I want to thank you all for attending our first earnings call and for starting this journey as a public company with us.

Our first question comes from Michael Ward of Citigroup.

Don, I wonder if you could talk a bit more about the ZF partnership that seems intriguing. How exactly is that going to work? Is it a supply relationship, development relationship? How is that going to work?

Thanks, Mike. The ZF relationship is primarily a supplier relationship. ZF, as you know, is one of the leading suppliers of steering columns and automotive components in general to the commercial trucking market. We have a long-standing relationship with them and use their components in our driverless trucks today. And this announcement further solidifies the partnership between Kodiak and ZF as we intend to scale our solution going into 2026.

That's great. It's a great company. Surajit, as you look at the fourth quarter, it seems like your annualized run rate at year-end is going to be somewhere in that $5 million range just from the Atlas relationship. Is that about right?

We expect meaningful growth in revenue. So Q4, as we have guided, we will be into...

As you exit, is that what we're talking about in that $5 million range for the annualized run rate?

Yes. I think that could be close to that number.

Okay. And then from what I can tell, your cash burn was about $35 million, you had some unusual there in 3Q. So about $35 million a quarter, that's about right?

In Q3, we had some high single-digit millions of one-time costs and public company-related costs. So we have guided for Q4 for free cash flow to be negative $36 million to negative $38 million.

Perfect. Okay. And just one last one, if I could, is I think you're on track to get to that 100-unit agreement with Atlas by the end of '26. That's your current target?

Yes, that's our target.

Our next question comes from the line of Andres Sheppard of Cantor Fitzgerald.

Congrats on all the great progress and congrats on the first public quarter. Certainly, incredible achievement. Don, I wanted to maybe touch on the long-haul operations that you're targeting to launch in the second half of next year. Can you maybe help us understand what is needed between now and then between that, I guess, 78% and 100%? Anything that you can point that we should be focusing on? And how confident are you in that target?

Thanks, Andres. Yes, we're excited to finally come out with some tracking metrics around our progress toward long haul. We've been talking about this for the last several quarters, and we wanted to be able to provide the market with some visibility and transparency into our development process. The ARM is effectively a measure of the material completeness of our safety case. And as you know, we don't launch a driverless product until our safety case is complete. Most of the work from here forward involves a lot of validation and testing of the system that includes simulation, that includes structured testing, and other forms of testing validation as we build up to that launch. And so we'll be providing more detail along the way. For now, this is our first data point as we progress, but you should expect to see quarter-over-quarter improvement as we move toward our expected driverless launch in the second half of 2026, and we feel reasonably confident that, that is an accurate timeline.

Wonderful. That's super helpful. I appreciate all that color. And maybe just as one follow-up. A question on liquidity. So with roughly $150 million in cash and equivalents now, how are you thinking about kind of capital needs going forward? I realize you're not guiding cash burn going forward, but how should we think about that liquidity and any potential additional capital needs?

Thanks for the question. This is Surajit here. I'll jump in here. As you see as a part of the de-SPAC transaction, which you just concluded, we had the largest capital raise in the history of the company, demonstrating our ability to raise across the capital structure and across several investors. And we feel excited about our momentum and the tremendous progress we are making in deploying our technology and scaling the business. So as we scale this business over both industrial and long haul, we should be able to also drive significant operating leverage and reduce our BOM costs. So we feel confident that as we execute on this strategy, we will be opportunistically seeking additional financings to strengthen our liquidity over the next 12 months, support the next phase of growth, and execute against our roadmap. So we feel confident we're raising that additional capital for the next few quarters.

Our next question comes from the line of James McIlree of Chardan Capital Markets.

When considering the entry into the long-haul market in the second half, do you need to reach 100% for some time before entering, or can you start with less than 100%? I'm interested in how you're using ARM as a factor for this market entry.

Thanks, Jim, for the question. There's no specific period of time required. It's more about meeting a minimum standard. We do need to achieve 100% on the readiness measure to feel confident that we have completed the safety case for launching our driverless product. However, there's no defined time frame between reaching 100% and actually conducting the first driverless run. It is still early to discuss specifics at such a detailed level. I would say we are aiming for the second half of next year. As we get closer to that point, we should be able to provide more clarity and certainty regarding the timing of our first deliveries.

That's very helpful. And then as far as customer additions are concerned, is it more likely that you enter the long-haul market before getting another industrial customer? Or is it the opposite that it's more likely to get another industrial customer first?

Well, we're dual tracking that. We're continuing to push to build our industrial business. As we talked about in our remarks earlier, we feel really good about growing that vertical. We have a great customer in Atlas, and this is really a crawl, walk, run approach to deploying autonomy. This is a safety-critical technology. We want to make sure that it's rightsized for our customers as our customers are learning with us. There's a lot in the product pillar that goes into actually efficiently deploying this product. And you can't just dump hundreds or thousands of vehicles on a customer overnight and expect them to be able to efficiently operate those vehicles to provide a useful benefit. And so there's a learning process as we go through this customer development, and that's really where we talked about the flywheel earlier on. And so we are looking to pick up additional industrial customers, and we'll have more on that as we continue to move through the quarters. At the same time, the team, especially the R&D team, the systems engineering team, and our validation team are working really hard to get the truck to the appropriate level of safety for deploying driverless. And so I think it's really a dual track multipronged effort. Those are parallel efforts. I think the goal would be to announce additional customers in both of those verticals along the way. I can't really say exactly to what time frame one would come before the other. These are both top issues.

Our next question comes from the line of Itay Michaeli of TD Cowen. Alright, we move to our next question, next question comes from the line of Mike Latimore of Northland Capital Markets.

Congrats on the first call here and doubling the units in the quarter. That's great. I think Nauto's score was very positive. Can you leverage that? Are you leveraging that for marketing purposes after new prospects? And can that be helpful even, I don't know, in keeping insurance costs down?

When discussing new technology, particularly in critical areas like automotive safety, building credibility and trust takes time. We believe that the results from Nauto highlight the safety of our system, especially in comparison to human drivers. This has been a longstanding question: how do these vehicles perform compared to human drivers, not only in terms of accident rates but also in driving behavior. What excites us is that these vehicles not only avoid accidents, which is a key metric, but they also drive responsibly and safely. This contributes to improved safety on our roads and helps alleviate traffic congestion. It’s crucial for us, as we expand this technology, to be responsible not just to our customers but also to the general public. These results demonstrate the trustworthiness of our system for those who may not see it in action. From a marketing angle, this also benefits Kodiak, especially when communicating with regulators who may not have firsthand experience but can see that our score represents over 1,000 fleets regarded as the safest on the road. We can present this data to our customers, showing that across our entire fleet we operate as safely, if not more safely, than human drivers. This underscores the credibility and reliability of our system overall.

Great. Obviously, a lot of focus on industrial and long haul. In the government vertical, a ton of focus just across the board on autonomous systems, whether it's in air, on sea, or on the ground. Can you give a little more context or color around opportunities you might see in the government vertical?

Yes. It's been tricky as of late with the government shutdown, as we mentioned in our remarks. We still remain convicted that the defense vertical is a large opportunity for autonomy. The Secretary of Defense actually just made some remarks emphasizing the importance of contested logistics and actually prioritizing commercial solutions within the Army and other branches' adoption of autonomy. We think this is the right approach, and we think that Kodiak has, we believe, the most mature, commercially viable autonomy solution that can be applied to defense applications. And of course, as you've seen, we've demonstrated that several times over across multiple vehicles in multiple different environments with our Ford F-150s and the versatility they bring, but also the Textron Systems RIPSAW platform, which is a fully threaded vehicle. We really demonstrated the ability to bring that commercial maturity into the defense market. And we believe that, that's what the defense market is ultimately looking for as they want to scale and productize this technology. So we remain very bullish in defense applications. It's an interesting world in the government space right now, but we think 2026 is going to be an exciting year. So I would say stay tuned for more.

Our next question comes from the line of Nancy Hipp.

This is Nancy Hipp on for Ravi Shanker. It would be helpful to hear a bit about how you're managing adverse weather, extreme environments or edge case scenarios in your autonomous system and how big a risk those are to scaling? I know you had a GenAI system to identify sort of these novel edge cases, but it would be helpful to hear more on that as you approach the on-highway launch in 2026.

Sure. In our industrial launch back in 2024, we delivered the first two driverless trucks in December of that year. We quickly got those trucks operating around the clock, which is important for our customer, Atlas, who operates 24/7. Additionally, these trucks have been handling adverse weather conditions for several quarters in a driverless manner. We expect to apply these capabilities to the highway environment when we launch our driverless product in the second half of next year. The team is actively working on validating these capabilities for highway use. This is all included in our safety case, which is currently at 78% on our readiness measure. Our AI system has already dealt with adverse weather conditions.

Got it. That's very helpful. And then maybe for my second question, it would be helpful to hear about feedback you're receiving from your current partners after the launch with Atlas. And sort of how do you see that decision-making cycle for customers to go from initial discussions to adopting driverless trucks into a pilot to eventually scaling?

It's been an interesting journey, and our perspective has shifted considerably over the last decade, especially leading up to the end of 2025. We believe customers and the market are truly enthusiastic about driverless deployment. It's more about when they can access it rather than being skeptical, which was not the case in the past. A few years back, many prospective carriers and truckload operators were doubtful. Nowadays, skepticism has diminished significantly. People are starting to understand that autonomy represents the future of transportation, not only in the commercial sector but also in private vehicles. We at Kodiak firmly believe automation will enhance the efficiency and safety of all transportation modes over time. Customers are eager to leverage this opportunity, recognizing the first-mover advantage, and they want to act swiftly. Our goal is to provide them with a safe and efficient solution that they can implement right away. This is where the flywheel effect comes into play. Currently, we have an industrial application up and running, which differs from highway operations, but the customer interactions are fundamentally similar to what we will bring to our highway customers when we eventually launch that product. These insights can only come from direct experience, and we believe this flywheel will speed up our journey as we begin to expand our highway deployment. Customers are excited and patiently awaiting the first driverless trucks, and we aim to be the leading provider of these solutions for them.

Our last question comes from the line of an indiscernible speaker.

On the routes in Dallas, as you demonstrate the safety case, Aurora faced some pushback from their partner, PACCAR, during their own safety validation process. I'm interested in knowing if there is a risk that you could prove the safety case, but the partner may disagree. Additionally, you mentioned J.B. Hunt. Do they have any influence in this situation? Do they have concerns? I would assume that if you're just transporting goods from one location to another, they shouldn't mind, but perhaps they don't want the brand associated with any potential risk. Can you explain who needs to approve, if anyone, for you to operate driverless on those trips from Dallas?

Thanks, Walter. It's a great question. I think there's like the legal sense of the question, who needs to legally sign off and then from a trustworthiness and good partnership perspective, there's who do you want to bring along. Our philosophy is we've always built our technology to be platform agnostic. We've shown that we can develop The Kodiak Driver and implement The Kodiak Driver across many different makes, models, and form factors of vehicles. This gives us flexibility. So we haven't announced the platform that we will be using for our initial highway deployment. I think you asked, is this a risk? Everything is a risk. I would definitely say it's a risk. At the same time, we think building the right relationships and the safety of the technology in the right way and bringing people along, including them in the process, is the right way to approach business. And so we think we have a path forward to deploy driverless vehicles without a driver and without an observer in the cab, and that's something that we definitely intend to do. But for sure, building trust with our partners is paramount in that process.

Sticking with the partnership question, you've chosen to enhance your capabilities, correct? You've generated $800,000 in revenue, while Aurora's revenue is around $1 million, indicating not much difference in the current quarter. I'm curious about the timeline for integration. Is it significant to be aligned off the production line? I understand it's still early, but should we not concern ourselves with this for an extended period? Or do you have any plans in mind for 2027 or 2028?

I don't believe it's necessary to set a specific date for a transition. The focus should be on the rollout and scale of our operations. Many people seem to think that we'll suddenly have thousands of autonomous trucks on the roads, but we don't believe that's realistic. The rollout will happen in stages, taking into account safety, efficiency, and operational factors. We are confident that our current approach can support thousands of trucks, which should be enough for the near to medium term. This also relies on the development timelines of our partners, OEMs, and other companies in the autonomy space, which we cannot control. I have emphasized for a long time that we don't want to be tied to other companies' schedules; we prefer to take control of our own path. Kodiak has done a great job in this regard, and we will maintain that philosophy in the coming years. We want to ensure we have a product ready for our customers when the time is right. As the ecosystem evolves and suppliers catch up, I believe we'll see broader opportunities, not just for Kodiak but for the entire industry. Therefore, I don't view it as a black-and-white situation or a specific deadline. Progress will happen gradually. We are collaborating closely with suppliers on both the Tier 1 and OEM sides, monitoring each other's advancements. This is not a major concern for us, and we believe the experience we've gained, especially with our upfit solution alongside our partner Roush, positions us well for success in the coming years.

Our last question comes from the line of Itay Michaeli of TD Cowen.

Can you hear me?

Yes, sir. We can.

Perfect. Sorry about before. Congrats on the first earnings call. So going back to the 78% long-haul arm, Don, I was hoping you could maybe share roughly where that metric was maybe 3, 6, 12 months ago. And then on the OTA that you did that reduced the remote assistance by over 50%, curious if you could talk a bit more about that as well and kind of what are some of the issues that were resolved with that update?

Yes, absolutely. Thanks for the question, Itay. I'm glad that we cleared up the mic issue. So we don't have any numbers, historical numbers to share, unfortunately. This is our first data point. And of course, we will share updated data points going into the future, so you can see the trends. So unfortunately, I don't have a number to share on the historical aspect of that. Obviously, over the first several quarters of the year, we were focused very hard on delivering additional driverless trucks to Atlas and really perfecting the operation of those vehicles in that environment. And as we turn our focus to highway and our highway customers over the course of 2026, we'll have a lot more updates for you as we go. In terms of the improvement, this is incredibly exciting because efficiency is ultimately what will drive margins. And while all autonomous vehicles today require some type of remote support in certain circumstances, remote assistance in certain circumstances. It is our job as R&D developers to drive down the moments that any kind of assistance is required. So there's no specific instances I can point to or specific cases. But you can imagine that these trucks are very conservative, and they often will come to a stop if they see something they're not sure about. There's a lot of potholes that are present in the Permian. And often our truck will stop and ask human assistance for confirmation that they can continue or should they drive around it or is it safe? And ultimately, we want that conservative behavior in our trucks. But as we improve the technology, as our AI improves, as our foundation model work improves, the scene understanding, and we gave several examples of these in our deck, our scene understanding improves dramatically. The trucks can start to handle those cases on their own, and they need to call for human support less and less. And so we've actually reduced that, as we said, over 50% in the last quarter, and that's a huge, huge win and a sign that the technology is accelerating very, very quickly, and we expect that type of acceleration to continue.

Thanks. It's a great question. Most of the capital expenditures are intended for future deployment as we require some lead time to acquire and assemble the AV hardware components. Thus, most of that is related to upcoming deployment and ramp, which I would describe as success-based. As we plan the deployment for each quarter, we usually make those purchases. However, it’s not always linear; at times, we might make bulk purchases if the pricing is favorable or if there are potential tariff issues. We take those factors into account as well.

Thank you. And ladies and gentlemen, this concludes Kodiak's Third Quarter 2025 Earnings Conference Call. Thank you for participating. You may now...